A couple of days ago, retired NHLer Martin St. Louis put out a tweet that reminded me of a topic that I had experimented with earlier in the spring.
My opinion- I use to love playing on the road because most of my coaches wouldn't worry about matching lines because they didn't have the last change-I always felt I played more on the road. Checkers on the road, chess at home! I like both but checkers is more fun! @NHL https://t.co/sARDNByDlp
— Martin St.Louis (@mstlouis_26) May 16, 2018
Now, as it turns out, St. Louis might be misremembering his full NHL career. According to Hockey-Reference, the Tampa Bay Lightning legend and Calgary Flames / New York Rangers occasional averaged 20:26 of ice time per game at home and 20:19 on the road; an insignificant difference with a slight tick in the opposite direction.
Breaking his career down into seasons, though, you’ll see that he averaged more ice time on the road in five of his first seven seasons, including his Hart, Pearson/Lindsay, Art Ross, and Stanley Cup winning seasons. Once he became a known superstar, though, the tables turned, and between 2006/07 and his last year in 2014/15, he only played more on the road just once: 2013/14, the year where he requested a trade and was eventually moved to the Rangers.
So maybe there’s a point to be made there: that his coaches weren’t likely to send him out to play “matchup” hockey when he was unproven, but leaned on him more and put him in tougher situations at home. After all, that’s one of the big things we hear about great coaches, right? That they know how to match lines, and really take advantage of last change?
https://twitter.com/JeffVeillette/status/963244414482239488
Moving away from St. Louis, and over to the thought I was already operating with: my “eye test” wasn’t overly happy with how Mike Babcock had matched players up at times. It felt like the team leaned far too hard on players that it felt to be defensively responsible, even when it was clear that those players; the Ron Hainsey, Roman Polak, Leo Komarov types, were out of gas or in over their heads. Pertaining to the Leafs’ players specifically, I wrote a piece back in March looking at how certain players were deployed in various score situations, and how their usage connected to Toronto’s record, and while I didn’t have a firm conclusion from it, it at least gave you an idea of what matchups or deployments the team was chasing.
With that in mind, I was curious to how much Babcock, a noted line matcher and tinkerer, benefited from having the last change at home. There’s no pure, objective way to do this, but my first hunch was to split possession into home and away segments and figure out the differences, and track it over the course of Babcock’s career. In my first go-around, I looked solely at Babcock, and I used Score-Adjusted Corsi as my metric. Here’s what I came up with for him:
The result was pretty shocking; in seven of Babcock’s last ten seasons, his road possession at 5-on-5 outweighed his home possession. His two worst years were this season and 2012/13. My suspicion is that the drop in Detroit coincides with the loss of Nicklas Lidstrom; the Hall of Famer played extremely tough competition until the very end, even as his average time on ice dropped, and that 2011-2013 fall, rise, and plummet seems to mimic Lidstrom’s one poor Relative Corsi season, his rebound in his last year, and his disappearance into retirement.
As for this year, Toronto’s attempts to play “survival” or “play to not lose” hockey in pockets likely did a lot of damage. Isolating these numbers to the “Its fun, but its dumb” era, a stretch of time that began after the Leafs’ first consensus loss to Ottawa on October 23rd and ended in time for the Next Century game against Carolina in mid-December, Toronto had a 46.8% Score-Adjusted Corsi at home and a 50.8% SAC on the road. Seeing that the agreed upon worst stretch of “over-coaching” for the Leafs this year showed a very, very clear impact, I decided to look at every Head Coach that was behind an NHL bench this year, and their history since 2007/08 (the first year that we have shot attempt data for).
I did this using two methods:
- Raw Corsi: No adjustments, just shot attempts. I figured that this should be the baseline, as it’s more than reasonable to expect a coach to have a mental note of how the flow of the game is going as its expecting, without expecting them to start weighing the flow based on the exact parameters of the scoreboard. That’s for the analysts after the fact. My feeling here is that if you’re getting beat out on the road by the raw numbers, you’re not just doing a below average job: you’ve gone straight up into bad. After all, you’d assume that just about every coach in the NHL can make a better plain-sight situation for themselves at home than on the road using their roster deployment.
- Score-Adjusted Corsi: Again, this should smooth out some variances that you’ll get if a team had a particularly good or bad homestand or road trip, which is necessary when an absolute maximum sample here for a given year is 41 games. I also imagine that coaches who come out in the negatives here would be a bit more interesting: a negative here would mean that these coaches are maybe not being aggressive enough in trying to get that next goal, rather than just straight up having no control over the flow of the game.
For this, I also included partial seasons (coaches who were hired/fired midway through), but researched the proper date ranges to make sure that there wasn’t any overlap. I did not include Peter Laviolette’s three-game season with the Flyers in 2013/14, and in the case of Bruce Boudreau in 2012 and Claude Julien in 2017, I included their stretches with both teams that employed them.
Here’s what I ended up with:
There’s a lot of interesting stuff here, far too much for me to parse and I’d definitely encourage someone to dig deeper. A few things of note.
- I’m a little sceptical of the names that are near the top and bottom that don’t have a lot of NHL coaching experience to work with here, but they seem to align with the trains a thought that the public eye has on them overall. Weight and Boughner names like Hakstol, Housley, Weight, and Boughner are already relatively quick having their feet to the fire, while Hynes, Cassidy, and Bednar have picked up some praise.
- Bill Peters and Jeff Blashill, on the other hand, are two guys that have received increasing heat from their fanbases (in Peters’ case, former fanbase, after his opt-out in Carolina and leap to Calgary), but are in the top third of the charts. They, coincidentally, are the two strongest examples of smaller-sample swings: both had strong first years with their teams (Peters +3.8 score-adjusted, Blashill +4.0) but declined to about even in their next year, and were flat out bad afterwards (Peters -1.4 & -1.7 in the last two years, Blashill -3.2 this year).
- I was curious to what effect John Hynes’ impact came from just double shifting Taylor Hall (which is the Occam’s Razor move that I would make if someone put me behind a bench, in total honesty) at home, but you instead see the opposite happening. Hall and 2017 top draft pick Nico Hischier see their 5v5 minutes go up at home, but play against weaker competition.The player who interested me in terms of getting leaned at home? 26-year-old Sophomore Blake Coleman. Coleman’s already picked up a bit of a buzz for being a takeaway wizard, check finisher, and shorthanded goal scorer, but its interesting to see how Hynes uses him when he gets last change; Coleman on the road gets an obscene amount of defensive zone starts, but that gets tempered a bit at home, with the caveat of him getting matched up against much tougher competition; the biggest spike on the team.
- With the above in mind, if someone were to take this idea to the player level: I’d imagine that you’d find the most interesting matchup players if you took guys who had low Zone Start ratios overall but started a noteworthy amount of additional shifts in the offensive zone at home. This would, presumably, find you the players who their coach believes enough in defensively to toss into a tough spot, without knowing who their opponent is (or the inverse for lopsided offensive players).
- Speaking of “just keep sending this guy over the boards and you’ll be fine” scenarios…
McLellan coached Joe Thornton for seven years and has coached Connor McDavid for three, and, when adjusting for the score has lost a chess match to a checker set in nine consecutive seasons. That’s legitimately impressive. - Randy Carlyle ranks pretty high on this list for a guy who gets ripped on as a coach on a frequent basis (you can sort of say the same thing about John Tortorella, but the gripes with him seem to be more ingrained in persona, with talent gripes being a mostly-temporary phase). That’s because he opened up the fancy stat era with the two best years on record, for this made-up stat, back-to-back:
Unadjusted, the Anaheim Ducks were an absurd 9.3% higher Corsi at home in 2007/08, and 6.3% better in 2008/09. Even with score adjustment, they were still remarkably better at home. How was that possible? Well, they had an immensely stacked team and a coach who was willing to be extremely aggressive with utilizing them:
The above table shows every Anaheim Ducks skater that played at least 500 minutes both at home and on the road between 2007/08 and 2008/09, and the difference between their home results and their road results.The famous shutdown line of Niedermayer, Pahlsson, and Moen from the great 2007 team? With Carlyle now able to pick their matchups, they get used in all three zones against tough opponents, rather than in the defensive zone as a constant hail mary. The defence gets a bit of a break, with the depth guys who don’t make this list getting some extra time. Perry, Getzlaf, and Selanne don’t get matched up as frequently against top-line talent. The result? All of the regulars feast a bit more. Okay, a lot more: outside of Selanne, every single person here sees a gigantic increase in possession, and, in the process, shot quality. - So where does it fall apart? Likely when most of the leave. The core starts to fall apart for cap reasons late in 2007/08, really comes off the rails towards the end of 2008/09, and by 2010, it’s basically the Getzlaf & Perry era at full tilt. Carlyle still ends up being an average matcher at that point, if not slightly above, but loses the huge edge he had with his perfectly-tailored, diamond encrusted toolbox. Carlyle’s methodology changes from here too: matchup results in recent years tend to lean towards gaining a surplus of shots for, rather than minimizing shots against.
- On that note, I did find it very interesting how much the differences in long-term Corsi For / Against aligned with what we know about a lot of these coaches “styles”. True to their reputation, a lot of the defensive-minded coaches get most of their matchup edge by decreasing shot attempts against, a lot of the run-and-gun guys gain them through increasing the frequency from their players, and the most mysterious, polarizing types seem to
If I were to go through every single coach to figure out what the tics were to the ups and downs of their match-up career, I’m sure there would be some very interesting results to be had. Maybe that’s something I revisit throughout the summer, or, regardless of my own plans, it’s definitely something I encourage any curious person who reads this to do.
Whether or not this methodology means anything in particular is a question far above my pay-grade. As well, matchups are just one facet of coaching ability; being good or bad according to this doesn’t necessarily mean that someone is a good or bad coach at this level.
Regardless, this was still a fun thought experiment to undertake, and one that leaves me more curious than when I started. Hopefully you can also find some takeaways from the above or use it to build your own, better ideas.
